Spatiotemporal dynamics of HIV infection

Mathematical models of the dynamics of infection with the human immunodeficiency virus (HIV) have contributed to tremendous advances over the past 20 years. This thesis extends this previous work by exploring the importance of spatial heterogeneity in HIV infection both in vitro and in vivo in patients treated with highly-active antiretroviral therapy.; Viral infections propagate locally in space, yet HIV infection has been widely regarded as equilibrated over the entire body of an infected patient. This dissertation constructs and explores a cellular automata model of viral spread at the cellular level. Coupling the automata to a blood compartment represented by a differential equation leads to a whole-body model of HIV infection that explicitly includes spatial effects at both the cellular and tissue levels.; These models are tested by comparison with experimental data. A central prediction of the spatial model is that, due to competition between Brownian motion and viral lability, HIV infectivity increases with target cell density. This production is verified in a series of in vitro experiments in cell culture. The predicted independence of inhibitory concentrations of antiretoviral agents is verified for nevirapine, but azidothymidine inhibits HIV replication less efficiently in more dense cultures. These in vitro results suggest that systems allowing cell concentrations closer to tissue densities would better reflect virus replication kinetics, although standard measures of relative drug susceptibility may accurately reflect in vivo conditions.; The coupled spatial model of in vivo dynamics is compared with novel mathematical analysis of experiments in HIV-infected patients. These analyses indicate that HIV DNA provides a useful marker of the size of long-lived cellular reservoirs of HIV. Levels of HIV DNA in peripheral blood are predictive of the average rate of residual virus production after years of treatment, regardless of whether patients initiate therapy during the primary or chronic stages of HIV infection. The nonlinear clearance of HIV DNA therefore predicts lifelong virus production, even in treated patients. Collectively, these results demonstrate that important features of the spatiotemporal dynamics of HIV infection both in vitro and in vivo are best explained with explicit spatial models.